Abstract Rationale The quality of medial tibial plateau (MT-Plateau) alignment is one of the key elements for accuracy and sensitivity to change of knee radiography in knee osteoarthritis (OA). Aim To evaluate the influence of the quality of the MT-Plateau alignment on the reproducibility of joint space width (JSW) measurement in knee radiographs. Methods One hundred and twenty-seven knee radiographs (99 OA), performed using a standardized radiographic procedure (Lyon schuss (LS) view). • Evaluation of the quality of MT-Plateau alignment. • Computerized measurement of the JSW, twice, 1-month apart, using a semi-automated and an automated method of measurement. • Assessment of the reproducibility of repeated measurements: calculation of intra-observer coefficient of correlation, smallest detectable difference (SDD) and coefficient of variation (CV). Results MT-Plateau alignment was satisfactory in 99 radiographs (77.9%). Reproducibility was excellent in both satisfactory and non-satisfactory radiographs, irrespective of the method of measurement used. The automated measurement was more reproducible than the semi-automated one (CV 1.15% and 3.23%). SDD and CV were better in satisfactory than in non-satisfactory MT-Plateau aligned radiographs. Conclusion These results confirm that computer measurement of the medial tibio-femoral JSW, from LS digitized radiographs, is highly reproducible, irrespective of the quality of the radiograph. However, the quality of the MT-Plateau alignment influences the reproducibility of JSW measurement. The automated measurement was more reproducible than the semi-automated one.